本研究收集坡地社區開發之邊坡破壞與非破壞案例，整理出影響坡地崩塌的因素，並配合崩塌因素評估準則表，將影響因素予以量化，利用多變量統計中之鑑別分析法和統計套裝軟體-SAS，對不同邊坡影響因素之變換組合以及降雨情況，做個別及整體性之分析，以初步建立坡地社區邊坡崩塌機率之預測模式。在分析過程中，除了可求取坡地社區崩塌之機率外，還可評估影響坡地崩塌因素之相對權重及彼此間相互影響的關係。由分析結果顯示，邊坡地質、人為因素及邊坡水文等三項因素具顯著鑑別力。本研究即依據此三項因素，分別就其主因素和子因素進行綜合和獨立分析，建立坡地社區邊坡崩塌機率之預測模式，各模式對樣本內和樣本外之預測均具有相當高之正確率。而且可根據坡地資料的取得完整程度，採用適當的評估方法。
To identify the influencing factors on landslide in hillside communities, several case histories with or without landslide are collected. These factors are quantified based on the criterion suggested in this research. By using the discriminant analysis of the multivariate statistical analysis method and the statistic software SAS, the probability of landslide of the hillside communities will be evaluated in terms of various combinations of influencing factors and weather conditions. Apart from the probability of landslide, the weighting of each individual factor and the interactive effects among the factors are also evaluated. Among the influencing factors for landslide, geological condition, human activities and drainage condition are the most discriminating factors as indicated by the SAS program. Based on these three factors and sub-factors, models to predict the probability of landslide are proposed. Different models can be chosen depending on the availability of field data. In general, the probability of landslide with high accuracy can be generated from the proposed models.
